Scalable, Automated ML Model Monitoring with KFServing and Hopsworks
summary
Describes an open-source streaming architecture that uses Kafka to capture model predictions, KFServing for serving models, and a Kubernetes operator for deploying components, with Spark Structured Streaming as the analysis engine. Presents an extendable monitoring framework built on Spark Structured Streaming to perform outlier detection and data-drift detection for continuous, scalable ML model monitoring and compares stream-processing benefits versus batch/isolated approaches.
AI-written summary · about
related talks
40mHow to leverage the Apache Kafka Ecosystem to productionize Machine Learning
37mBig Data Analytics, Machine Learning, Real Time Processing
26mOptimal Approaches for Real-Time Machine Learning with Apache Spark on Kubernetes.
16mAnalysing Real time Activity Streams for Security
30mDeploy end to end ML pipeline using Apache spark ...
42mHow to Leverage the Apache Kafka Ecosystem to...
41mStreams to the River: Powering an Online XGBoost Classifier with Kafka
33mDeploying and Scaling Spark ML and Tensorflow AI Models